70 platform-agnostic checks from test design through launch monitoring — SRM, guardrails, snippet install, goals, targeting, preview, and statistical validity. Works with any A/B testing tool.
GrowthLayer and enterprise CRO post-mortems show most failed tests break down here — weak hypotheses, wrong metrics, or unrealistic sample plans. Pass every item before a single visitor enters the experiment.
Verify your testing tool is installed correctly so visitors bucket reliably and conversions are captured from the first pageview.
Flicker and slow loads don't just hurt UX — they bias test results when users leave before the variant renders.
A perfectly executed variant is worthless if conversions aren't counted accurately. Validate tracking before traffic, not after.
Confirm the right visitors enter the test at the intended split — misconfigured targeting silently wastes weeks of traffic.
Validate every variant against the hypothesis before launch. Brillmark's QA process treats this as the final gate — one broken variant invalidates the entire experiment.
The first 24–48 hours catch Sample Ratio Mismatch, tracking failures, and traffic skew — issues that make results uninterpretable if ignored.
Protect statistical validity when reading results and close the loop so learnings compound — even flat and losing tests have value.
Works with any A/B testing platform. Plan smarter experiments and ship with confidence.
Get FREE AccessSample size calculator, duration calculator, and significance calculator help you plan and validate results alongside this checklist.
Progress saves automatically in your browser. Use Export and Import in the toolbar to back up or share with your team.
Use this interactive checklist before and after every experiment — no spreadsheet required. Covers test design, snippet install, anti-flicker, goals, targeting, preview QA, launch monitoring, and statistical validity.
Yes. All 70 checks are free with no email gate. Works with any client-side A/B testing platform.
SRM means your actual traffic split doesn't match your configured split — often from a tracking or bucketing bug. Check within 24 hours of launch; invalid SRM means results can't be trusted.
Run through Test Setup and Snippet sections before launch, Preview QA the day before go-live, then Launch Monitoring daily for the first 48 hours. Use After Test when the experiment ends.
